Evaluation of the OSC-TV Reconstruction Algorithm for Optical Cone-Beam Computed Tomography

Abstract

This paper evaluates an iterative reconstruction technique, the OSC-TV regularized maximum-likelihood algorithm, in the field of optical cone-beam computed tomography (CT). This imaging modality is commonly used in gel-based three-dimensional (3D) dosimetry for external beam radiotherapy. It is conjectured that model-based iterative reconstruction may improve optical CT image quality and encourage a wider use of this modality in clinical dosimetry. The OSC-TV algorithm was evaluated using experimental data from a commercially available optical cone-beam CT system, and the resulting images were compared with those obtained via filtered backprojection. Physical phantoms were scanned with the goal of evaluating the modulation transfer function, reconstruction precision and accuracy. Reconstruction artifacts were studied as well. The OSC-TV algorithm yielded images of improved quality compared to the filtered backprojections, thanks to a model-based simulation of the photon attenuation process. It was shown to improve spatial resolution, reduce noise and improve accuracy and precision of linear attenuation coefficients’ estimation. In addition, image artifacts due to acquisition process and total signal loss were corrected. Overall, this study demonstrates that OSC-TV represents a valuable alternative to filtered backprojection for optical cone-beam CT.